9/13/00 EIIP Virtual Forum Transcript: the Hurricane Floyd Evacuation in South Carolina
Edited Version of September 13, 2000 Transcript
EIIP Virtual Forum Presentation
"Learning from the Hurricane Floyd Evacuation in South Carolina"
Dr. Kirstin Dow
University of South Carolina
Avagene Moore
EIIP Coordinator
The original unedited transcript of the September 13, 2000 online Virtual Library presentation is available in the EIIP Virtual Library Archives ( The following version of the transcript has been edited for easier reading and comprehension. Typos were corrected, date/time/names attributed by the software to each input were deleted but the content of questions and responses are as stated by each participant. Answers to participants’ questions are grouped beneath the appropriate question to facilitate meaning.
[Opening]
- Avagene Moore: Welcome to the EIIP Virtual Library!
Our discussion today is based on a Natural Hazards Center Quick Response Report (QRR # 128) titled "South Carolina's Response to Hurricane Floyd," co-authored by Dr. Kirstin Dow and Dr. Susan Cutter. We are pleased to have Dr. Kirstin Dow with us today to share her research with us. Kirstin was with us in July '98 to discuss "Crying Wolf: Repeat Response to Hurricane Evacuation Orders," her earlier research on factors influencing evacuation decisions. See .
Kirstin is an Assistant Professor of Geography at the University of South Carolina. Her research interests center on issues of vulnerability to environmental hazards. Currently she is involved in research projects on urban ecosystems and hazards and vulnerability in urban areas. And now, please help me welcome our speaker, Dr. Kirstin Dow. After Kirstin presents her paper, I will remind you of the protocol for the Q&A segment of today's session. Kirstin, I turn the floor to you now.
[Presentation]
Kirstin Dow: Hello, thank you. I am very pleased to be invited back. Today I am going to talk about new research and research that builds on the 1998 work.
Learning from the Hurricane Floyd Evacuation in South Carolina
Over the past 4 years six hurricanes of varying strength have threatened the South Carolina coast. When Hurricane Hugo, a category 4 hurricane and the worst storm in the state in recent history made landfall in 1989, over 60% of the population evacuated (Baker 1994). Nonetheless, the evacuation for Hurricane Floyd was unprecedented in this state. Approximately ½ million people were involved in the evacuation that resulted in bumper-to-bumper traffic on the interstate highways. Under these conditions, a distance of 100 miles sometimes took over 15 hours by car. Criticism mounted quickly as people stuck in traffic picked up their cell phones and radio stations broadcast awareness of the problems.
Not until late in the day of the first wave of evacuation did the State agencies succeed in transforming a major stretch of Interstate 26 between Charleston and Columbia into a one-way route away from the coast. All evacuees were off the roads by the time Floyd closed in on the Carolinas. Nevertheless, the evacuation process came under heavy scrutiny and the "success" of the process occupied the center of that debate.
I will comment on three primary issues:
1) Response to mandatory evacuation, including consideration of the potential "crying wolf" syndrome from previous storms;
2) Longitudinal data on response in Horry County; and
3) Traffic on the interstate and non-use of alternate routes
Most of the data I will discuss was gathered in 2 surveys:
1) Statewide survey of residents in affected areas;
2) Re-survey of Horry County residents
The following statistics describe the Coastal Carolina Survey. I will introduce more information on the other survey later when I talk about Horry County, specifically.
* Phone interviews of coastal residents October 25-November 9, 1999;
* Six weeks after Floyd, 513 completed interviews, 19 partial;
* Response rate 63.5% * Sampling error 4.2%;* Conducted by USC’s Institute of Public Affairs Survey Research Lab
Findings of the Coastal Carolina Survey:
Sixty-four percent of respondents, sampled from among residents of the mandatory evacuation area, evacuated in anticipation of Hurricane Floyd. This is the highest coastal average reported for the state since Hurricane Hugo, a 1989 category 4 storm.
Evacuation Statistics within the Mandatory Evacuation Zone:
* Evacuated their home 64% (Left the area 59%; Stayed in the area 5%)
* Tried to evacuate, but could not 2%;
* Did not evacuate 34%
These are generally consistent with findings of the broader East Coast study (Baker 2000).
While the severity of the storm was a top concern all along the coast, landfall location emerged as more important in the northern part of the state, especially in Horry County. Throughout the coastal areas, the perceived safety of the home was the major factor in residents deciding to stay in an area despite evacuation orders.
Statistics on the Most Important Factor in Decisions to Evacuate (First response given)
* Severity of the storm-23%;
* Past experience with hurricanes-17%;
* Governor's order-17%
If we look at all responses, we see the same ranking of issues.
Statistics on decision to evacuate: All responses (Ranked 1,2, and 3 in importance)
* Severity of the storm-20%;
* Past experience-17%;
* Governor's order-13%;
* Probability of landfall in the area-9%
Throughout the coastal areas, the perceived safety of the home was the major factor in residents deciding to stay in an area despite evacuation orders.
These are the top three Important Factors in Decisions to Stay
1) Felt safer at home-16%
2) Probability of direct hit-11%; * Not want to leave pets-11%; * too much traffic-11%
3) Info from the Weather Channel-7%; * Work responsibilities-7%
Concerns over the volume of traffic were tied among the second most commonly mentioned factor as 11% of non-evacuees included this concern. In the past 4 years of research, this is the first time that traffic concerns have been among the top three reasons to stay rather than evacuate. But traffic has been emerging as a concern. Horry County evacuees for Hurricane Bonnie (1998) also mentioned traffic as a reason for non-evacuation.
Given the diversity of concerns and the 35% non-evacuation rate, we asked respondents two questions about their support for calling evacuations.
1) A. Given what you knew about the storm before it made landfall, was evacuation the proper response to Hurricane Floyd?
2) B. On another issue: Given the uncertainties about the hurricane track, was evacuation the appropriate response to Hurricane Floyd?
86% said yes to question A; 84% said Yes to question B. Despite decisions to stay and importance of non-governmental sources in decision-making respondents strongly supported precautionary approach to calling for evacuation. We have a longitudinal data set for a group of Horry County residents that covers 6 hurricane evacuations and 4 years. While we are on the topic of why people evacuate, I would like to say a bit more about these findings.
Survey Description.
Many respondents to a 1998 study of the Hurricane Bonnie evacuation agreed to participate in follow-up interviews. The 1998 research took place in September and drew on a convenient sample of Horry County residents for a 10-minute face-to-face survey. At that time, residents were asked their evacuation decisions for Hurricane Bonnie, as well as the Hurricanes Bertha and Fran, which occurred two years earlier.
Horry County:Selected Respondents
* Sample of previously-contacted respondents (N=123);
* Not statistically representative;
* Almost all were in the evacuated area (95%).
[SLIDE 1]
In reviewing Horry County Evacuation History, evacuation rates appear to be more consistent with the maximum wind speeds than landfall strength. This relationship is also noted in a longitudinal study conducted by Whitehead et al. (2000) of evacuation of North Carolina residents for Hurricanes Bonnie, Dennis, and Floyd. Their study found that evacuation rates are correlated with storm strength. In comparison with Hurricane Bonnie evacuation rates, NC residents were more likely to evacuate for Floyd (p=0.01) and less likely to evacuate for Dennis (p=0.01) (Whitehead et al. 2000).
As our data on Hurricanes Bertha and Bonnie illustrate, the relationship to storm intensity does not hold in all situations. Wind speed is only one dimension of the risk posed; flood potential and probability of nearby landfall are also frequently mentioned risk issues.
Finally, considered as a temporal sequence, the relatively small changes in evacuation rates for Hurricanes Bertha, Bonnie and Fran generally seem to support more detailed analyses indicating that the so-called "crying wolf" does not have a major impact on evacuation rates.
[SLIDE 2]
We have reasons for evacuation or staying only for the 1998 and 1999 hurricanes. While a correlation with storm strength is indicated, this small sample suggests that the reasons for evacuation are shifting in importance. The safety of the house becomes more important in more intense storms as does the role of past experience.
On another point, the evacuation actions of Horry County residents in response to Hurricane Floyd is fairly consistent with the decision criteria they setout for themselves in 1998. In 1998 following Hurricane Bonnie, we asked these residents what they would do if another hurricane threatened the South Carolina coast. We compared this to their actual behavior in Hurricane Floyd.
Approximately 60% of respondents said that their decision would "depend" on various factors. One group responded that their decision to evacuate would "depend on" specified decision criteria. To a large extent, at least 75%, this group as they expected they would taking into the account the strength of the storm, the relative proximity of the anticipated landfall, and the governor's evacuation order.
Among the group that replied, "yes" they would evacuate in the future, 97% did so for Floyd. However, the group that said "no," they would not evacuate in the future showed the greatest deviation from plans in response to Hurricane Floyd, 70% of them evacuated.
Only 12 of the 19 "No future evacuations" respondents explained what convinced them to evacuate for Floyd; of that group 58% indicated concern over the strength of the storm and probability of nearby landfall.
Chi-square analysis of these data provide strong evidence of a relationship (p=0.000) between anticipated and actual evacuations, despite the changes in opinions of the respondents who had not planned to evacuate.
These Horry County residents are also actively seeking information from a wide variety of sources and relying on considerable amounts of past experience.
Statistics on Experience:
63% Have been advised to evacuate over 5 times;
19% Have been advised to evacuate over 10 times;
44% Have suffered hurricane related damage
Statistics on Information Gathering:
58% Leave a news source on all day once they are aware that a hurricane is within 2-3 days of the East Coast and might threaten South Carolina;
53% Leave a news source on all day once they are aware that a hurricane is within 2-3 days of the East Coast and might threaten South Carolina;
30% Consult the internet for information
These longitudinal surveys indicate that some residents have developed fairly robust strategies for making evacuation decisions that are more dependent on individuals' household circumstances, assessments, and preferences than official orders. The strategies that they are using for making decisions about evacuation are drawing on a more diverse and individualized set of criteria than has been identified in previous surveys. Making these decisions involves diligent monitoring of a variety of information sources and evaluation of past experiences. We have named this emerging subset of coastal residents as "hurricane savvy" based on these characteristics of stable strategies for response, active learning and investigation of hurricane information, and high levels of experience. With that look at why people evacuate, I want to return to the general population survey and issues of traffic once people evacuate.
Attempts to account for that traffic volume after the fact identified several contributing factors in addition to the distance traveled. It is clear that most research has been done on adequacy of sheltering destinations not the characteristics of the trip that influence levels of demand on parts of the transportation infrastructure.
Informal observation generated many suggestions of the sources of concentrated demand.
Because traffic was concentrated on the Interstate highways and was comparatively light on the state and county roads, some observers speculated that evacuees did not have maps with them and were reluctant to take alternative, less familiar routes. Others noted that traffic was worse because some households took more than one car. Finally, the relatively short interval between the announcement of the voluntary and mandatory evacuations, as well as the evacuees traveling north from Georgia and Florida, contributed to the greater congestion.
The distances traveled did not receive a great deal of press attention and did not receive a great deal of direct attention, but we included it as a related issue. Our questions addressed each of these topics.
According to our survey of coastal Carolina residents, evacuees relied heavily on the Interstate system while traffic on smaller highways and roads was reportedly much lighter.
* Interstate-26, between Charleston and Columbia, was the most heavily used route during the evacuation with 19% of respondents taking it during part of their evacuation journey.
* Interstate 95, the second most commonly used route, carried about 10% of the respondents.
* In addition, I-95 carried considerable traffic from Florida and Georgia.
This preference for the Interstate seems to be related to more than lack of information about alternatives. Approximately 65% of respondents had maps in their cars. However, only about half of respondents with maps used those maps to select their evacuation routes.
Anecdotal information gathered by listening to radio call-in conducted during the evacuation suggests that navigational instructions are only one dimension of the route information evacuees are seeking. With 50% of the evacuees having easy access to information on the locations of alternative routes and not using it, more work is needed on this topic.
The traffic jam was also blamed on the heavy concentration of people on the roads due to the close spacing of the Governor's voluntary and mandatory evacuation orders. The voluntary evacuation was called at 7 am Tuesday, September 14, 1999 and the mandatory evacuation order followed at 5 hours later (noon).
[SLIDE 3]
The majority of residents, 61%, left on Tuesday between 6 am and 9 p.m. followed by a second large group of 31% during the daylight hours of Wednesday, 15th. Only a small percentage left Monday (5%) prior to any advisories or evacuation orders. In other words, an estimated 359,000- 412,000 people left the coast on Tuesday. The vast majority left in one of two periods, either between 9 am and noon (25.4%) or from noon to 3 p.m. (22.6%). The tendency to leave during daylight hours and to show some clustering around the time of the evacuation order is largely consistent with other hurricane evacuation reports.
Average distances traveled to evacuate appear to have increased. Of the evacuees among the coastal Carolina respondents, 56% left South Carolina, 32% stayed in state, and 9% stayed in their county of origin for Hurricane Floyd. The number of evacuees who traveled out of state is significantly higher than reported in past evacuations. The percentage of respondents evacuating within their local areas is lower than observed in past studies of hurricane evacuation in South Carolina. (The third option here would be to evacuate in state).
In contrast, past surveys of South Carolina residents who evacuated from Hurricanes Bertha and Fran in 1996 found that 15% and 28% respectively, evacuated out of state (Baker 1985; Baker 1997). Evacuation in the local area was more common in these hurricane earlier events with approximately 50% evacuating in of the "local" area for Diana in 1984, and 22% or Bertha and 19% for Fran in 1996 evacuating in their own neighborhood (Baker 1985; Baker 1997). Evacuees from Hurricane Floyd clearly traveled longer distances.
[SLIDE 4]
Slide 4 shows the destinations of respondents broken down by according to densely populated counties running from South to North along the coast. All nearby states, including the more distant Tennessee, were destinations. Baker (2000) and Gladwin (1999) report significant levels of out of state travel for residents of other states evacuating from Floyd.
It is also important to note that some evacuees traveled to North Carolina moving into areas also at risk due to the uncertainties in landfall location. In this case, North Carolina received both the impacts of landfall and much of the heavy rainfall and associated flooding, resulting in the greatest disaster in that state's history.
[SLIDE 5]
For those who evacuated in South Carolina, the larger inland cities -- Columbia, Augusta, and Greenville -- were generally important destinations. Slide 5 indicates another dimension of the regional variability of evacuation practices among residents in the state.
None of the evacuees in the southernmost part of the state (Beaufort) stayed in their counties of origin. While in the Charleston area and particularly in the Horry County/Myrtle Beach regions, more residents evacuated to other destinations within their home county. These regional variations in evacuation travel will need greater attention to support future evacuation transportation planning. Time to sum this up.
Conclusions
* Evacuation rate of 64.2% (+/- 4.2%) in coastal SC evacuation zone (Roughly 590,750 to 673,455 residents left SC coastal counties in anticipation of Hurricane Floyd).
* Residents are traveling further in their evacuation than indicated in past studies.
* In Horry County, robust, active decision strategies to cope with hurricanes evacuations characterize a hurricane-savvy population.
In closing, I think we learned some lessons and identified some areas where we need to concentrate learning efforts:
* Public uses information other than evacuation advisories/orders in decision-making.
* Responding to "hurricane savvy" groups may require targeted risk communication to pay more attention to issues of preparedness and the role of experience in guiding evacuation decisions.
* There is need for improvement in navigational aids and reduction in travel time.